Kβ Valence to Core X-ray Emission Studies of Cu(I) Binding Proteins with Mixed Methionine – Histidine Coordination. Relevance to the Reactivity of the M- and H-sites of Peptidylglycine Monooxygenase

Abstract

Biological systems use copper as a redox center in many metalloproteins, where the role of the metal is to cycle between its +1 and +2 oxidation states. This chemistry requires the redox potential to be in a range that can stabilize both Cu(I) and Cu(II) states, and often involves protein-derived ligand sets involving mixed histidine-methionine coordination that balance the preferences of both oxidation states. Transport proteins, on the other hand, utilize copper in the Cu(I) state, and often contain sites comprised predominately of the cuprophilic residue methionine. The electronic factors that allow enzymes and transporters to balance their redox requirements are complex, and are often elusive due to the dearth of spectroscopic probes of the Cu(I) state. Here we present the novel application of X-ray emission spectroscopy to copper proteins via a study of a series of mixed His - Met copper sites where the ligand set varies in a systematic way between the His₃ and Met₃ limits. The sites are derived from the wild-type peptidylglycine monooxygenase (PHM), two single-site variants which replicate each of its two copper sites (Cu_M-site and Cu_H-site), and the transporters CusF and CusB. Clear differences are observed in the Kβ2,5 region at the Met₃ and His₃ limits. CusB (Met₃) has a distinct peak at 8978.4 eV with a broad shoulder at 8975.6 eV, whereas Cu_H (His₃) has two well-resolved features: a more intense feature at 8974.8 eV and a second at 8977.2 eV. The mixed coordination sphere CusF (Met₂His) and the PHM Cu_M variant (Met₁His₂) have very similar spectra consisting of two features at 8975.2 eV and 8977.8 eV Analysis of DFT calculated spectra indicate that the intensity of the higher energy peak near 8978 eV is mediated by mixing of ligand-based orbitals into the Cu d¹⁰ manifold, with S from Met providing more intensity by facilitating increased Cu p-d mixing. Furthermore, reaction of WT PHM with CO (an oxygen analogue) produced the M-site CO complex, which showed a unique XES spectrum that could be computationally reproduced by including interactions between Cu(I) and the CO ligand. The study suggests that the valence-to-core (VtC) region can serve as a probe of not only ligand speciation, but also offer insight into the coordination geometry, in a fashion similar to XAS pre-edges, and may be sufficiently sensitive to the coordination of exogenous ligands to be useful in the study of reaction mechanisms.

Introduction

Copper metal centers are present in a variety of metalloproteins where they serve as redox centers, often cycling between the physiologically accessible +1 and +2 oxidation states. Oxidases bind molecular oxygen at Cu(I) sites and convert it to “activated” reduced forms, such as superoxo or peroxo species, which are reactive towards organic substrates. Electron transfer proteins utilize the two oxidation states to shuttle electrons from reductants into these catalytic sites. This chemistry requires a fine balance of coordinating ligands in order to set the redox potential at an appropriate level. Often a combination of histidine and methionine residues are utilized, which leverage the cupriphilic (Cu(II)) and cuprophilic (Cu(I)) properties of histidine and methionine, respectively. Separate systems have evolved to sequester and transport copper within the cell, and here the reduced Cu(I) state is preferred owing to the reducing environment of the cytoplasm. Predictably, these transport proteins favor Met ligation over His, but often still exhibit mixed Met - His coordination at their Cu(I) binding sites. The detailed electronic factors that contribute to the stability and reactivity of Met/His ligand sets are important yet elusive factors underpinning the cellular chemistry of copper, due largely to a lack of experimental probes of the Cu(I) state. Here we use X-ray emission spectroscopy (XES) to study the ground state properties of a series of Met/His copper binding sites derived from (a) the enzyme peptidylglycine monooxygenase and (b) the transporters CusF and CusB.

Peptidylglycine α-hydroxylating monooxygenase (PHM) catalyzes stereospecific α-C hydroxylation of C-terminal glycines, the first step in the α-amidation of hormones, growth factors and neurotransmitters^1,2 The molecular oxygen-dependent reaction requires two equivalents of ascorbate as exogenous reductant, releasing water and semidehydroascorbate as byproducts. Crystallographic characterization reveals two distinct Cu centers in the protein active site separated by 11 Å.^3,4 The Cu_H site, believed to serve primarily as an electron transfer site, is coordinated by three histidines (His; H107, H108, H172) in a roughly “T-shaped” geometry. The Cu_M site, where oxygen binding and hydroxylation occur⁵, has a mixed coordination sphere consisting of two histidines (H242, H244) and a methionine (Met; M314), (Figure 1). Spectroscopic studies on the cupric and cuprous forms of the enzyme have provided oxidation-state specific structural information.^6–10 In the oxidized form the Cu_M-center is bound by two His residues and two water molecules in what appears to be tetragonal geometry with a long axial bond to the Met residue, while on reduction the water ligands dissociate and the thioether of M314 moves closer to the copper. For the Cu_H site, the oxidized structure coordinates a water ligand in addition to its three His residues to complete the 4-coordination expected for a Cu(II) center, and on reduction this water is again lost. Exogenous ligands (O₂, CO, peroxide, azide) bind to the catalytic Cu_M site but appear to be excluded from the electron transfer Cu_H site by electronic or steric factors which are not well understood^5,10–13.

Figure 1.

(a) WT PHM metal binding sites ((PDB: 1PHM). (b) CusF metal binding site (PDB:2VB2), sulfur (yellow), nitrogen (blue), copper (orange). (c) in silico CusB metal binding site (adapted from reference 23: “Reprinted with permission Ucisik, M. N.; Chakravorty, D. K.; Merz, K. M., Jr. Biochemistry 2013, 52, 6911. Copyright 2013 American Chemical Society.”

The cus operon of E. coli encodes four structural genes – CBAF – where CusCBA forms a tripartite complex that spans the periplasmic space, and imparts resistance to both Ag(I) and Cu(I)^14–16. The metallo-sites of the components of the Cus system have rich coordination chemistry, which is dominated by methionine ligation. CusF, which serves as a periplasmic metallochaperone, coordinates a single Cu(I) or Ag(I) ion in a (Met)₂His environment but the site is capped by a unique Π-cation interaction with a tryptophan residue (Fig. 1b).^17–19 CusB, a periplasmic membrane fusion protein, contains three conserved Met residues (M21, M36, M38) in its disordered N-terminal domain, which have been implicated as the binding site for Cu(I) and Ag(I) by XAS (Fig. 1c).^20–22 DFT and QM/MM calculations have provided further insight through in silico structures of the CusB N-terminus in both apo- and metal-bound forms,²³ where metal binding appears to induce a significant structural rearrangement suggestive of a function involving a metal-induced conformational switch. The CusA metal binding site, as determined from crystal soaks, lies within a deep cleft in the periplasmic domain and is suggested to consist of three Met residues, although other potential ligands (particularly E625) are within coordinating distance ^24,25. Completing the tripartite complex, an outer membrane channel is proposed to be formed by a trimeric assembly of CusC.²⁶ Recent multi-edge XAS studies have firmly established the individual roles of the protein components of the transporter.²⁷ CusF is able to sense the periplasmic metal load, and under high flux transfers Cu(I)/Ag(I) to CusB, generating an active CusB conformer which binds to CusA and activates the pump. The activated form of CusA can now also accept metal from the CusF chaperone, and transport it out of the cell. However, as the periplasmic metal flux falls, back-transfer from B to F leads to the apo-form of CusB which can no longer interact with CusA and shuts off further transport. For the current study, the Met₂His and Met₃ ligand sets of CusF and CusB, respectively, are uniquely suitable for studying the electronic interactions of the d¹⁰ Cu(I) in mixed thioether and imidazole ligand environments.

To understand the role of His and Met residues in copper-based catalytic and transport systems in more detail, Kβ valence-to-core (VtC) XES was applied as a direct probe of ligand coordination environment and electronic structure. Kβ VtC XES is due to transitions from valence electrons to the metal 1s core-hole.^28,29 This method has previously been used to identify the presence of a bridging oxygen atom in the OEC, the protonation state of oxo bridged Mn(IV) dimers and to identify the presence of a central carbon in nitrogenase.^30–33 Data were collected and analyzed for WT PHM and two recently characterized variants in which only the Cu_M-site (H107AH108A) or only the Cu_H-site (H242A) were occupied, providing His₂Met and His₃ isolated ligand sets,⁹ and on wild type PHM with CO bound to the Cu_M site (PHM-CO)¹² as a model for the PHM-dioxygen catalytic complex. These systems were compared with CusF (Met₂His) and CusB (Met₃) providing a systematic series of Cu(I) coordination spheres ranging from His₃ to Met₃ to facilitate spectral analysis. Correlation of the experimental spectra with theoretical calculations allowed the comparison of the Cu-ligand binding interaction for methionine versus histidine, determined contributions to the spectra from metal d-orbitals, and showed how these contributions are affected by ligand orientation and binding mode. Lastly, the sensitivity of VtC XES to detect small molecule binding in PHM was investigated, as a prelude to future mechanistic studies. To the best of our knowledge, this represents the first VtC XES application to copper proteins, however, previous studies have shown the potential of Cu VtC XES by exploring ligand protonation state in Cu(II) models of galactose oxidase.³⁴

Experimental Section

Protein Expression and Sample Preparation

Preparation of PHM samples

WT PHM and its single site variants were prepared as described previously⁹. Protein expression was carried out in hollow fiber bioreactors^6,35,36 as follows. The stably transfected cell lines were thawed from freezer stock into a T75 flask with 20mL of DMEM/F12 medium containing 10% FCII serum (Fisher). At 80% confluence, the cells were passed into five NUNC triple flasks (500cm² per flask), which were also grown to confluence. The cells were trypsinized and resuspended in 50 mL DMEM/F12 medium with 10% FCII serum prior to inoculation into the extra-capillary space (ECS) of a Hollow Fiber Bioreactor (Fibercell Systems 4300-C2008, MWCO 5kD, 3000cm² surface area) precultured with 2L of 50mM PBS pH 7.35 and 2L of DMEM/F12 10% FCII serum.^7,12,35 Individual bioreactors containing each of the variants were fed with DMEM/F12/10% FCII serum for a month. The serum level was then reduced to 0.5%, at which point spent medium (20mL) from the ECS was collected every other day and frozen at −20°C for later purification. About a month's worth of the bioreactor harvest (300mL) for each variant was purified as previously described.^7,12,35 Copper reconstitution was carried out as follows. For WT PHM, purified enzyme was dialyzed against 20mM sodium phosphate buffer, pH 8.0 and then reconstituted with 2.5 mole equivalents Cu(II) sulfate per protein followed by two cycles of dialysis to remove unbound cupric ions. For the single-site variants (H107108A and H242A), the purified protein was initially dialyzed against 20mM sodium phosphate buffer, pH 8.0 overnight, reconstituted with 2.5 equivalents of Cu(II) sulfate using a syringe pump, at a rate of 60ul/hr, followed by exhaustive dialysis against copper-free phosphate buffer at the same pH and ionic strength. This procedure resulted in copper to protein ratios close to 1. Thereafter, the single-site mutants were reconstituted with 1.3 equivalents of Cu(II) sulfate, and dialyzed overnight against 20mM sodium phosphate buffer, pH 8. The copper concentrations were determined using a Perkin-Elmer Optima 2000 DV inductively coupled plasma optical emission spectrometer (ICP-OES). Protein concentration was determined by on a Cary-50 UV-vis spectrophotometer at room temperature using an extinction coefficient for a 1% solution at 280nm of 0.980.

Reduced (Cu(I)-containing) derivatives were generated by reduction with a 5-fold excess of ascorbate buffered at the same pH as the protein sample. The protocol was carried out under anaerobic conditions to avoid air oxidation. The carbon monoxide derivative of the reduced enzyme was prepared by first vacuum flushing the sample with argon and then incubating under 1 atm of CO gas for 15 minutes. All samples were flash-frozen in 2mm × 10 mm lucite cuvettes immediately after preparation.

Preparation of CusF and CusB

The study utilized a variant of CusF missing its first 5 amino acid residues (CusF_6-88, hereafter termed CusF) and an N-terminal truncation variant of CusB (CusB-NT_1–61, hereafter termed CusB). E. coli BL21 (λDE3) cells containing the CusF_6-88-trx-his6-tev plasmid were grown from a freshly streaked plate into LB media containing 100 μg/mL ampicillin at 37 °C until they reached an OD₆₀₀ of 0.8, at which point protein expression was induced with 0.4 mM of isopropyl β-D-1-thiogalactopyranoside (IPTG). Growth was continued at 37 °C for 4 hours, after which the cells were harvested by centrifugation and pelleted. The cells were resuspended, lysed using the French pressure method, and centrifuged to remove cell debris. The filtered supernatant was poured over a Ni-NTA resin column, rinsed with buffer, and eluted using a 250 mM imidazole buffer rinse. To remove the His6-Trx tag, tobacco etch virus (TeV) protease and 5 mM β-mercaptoethanol was added to the protein solution and the mixture incubated at 20 °C overnight. After dialysis the protein solution was repurified on a Ni-NTA resin column, which removed the cleaved His-tag to yield pure apo CusF. The CusB-NT protein was purified from the CusB-NT-trx-his6-tev plasmid in the same manner as for CusF.

Cu(I) forms of CusF and CusB-NT, were prepared as follows. Protein concentration was determined by the BCA method, then aliquots of appropriate concentration and volume were kept chilled overnight in a Coy anaerobic chamber to give anaerobic protein. Tetrakis(acetonitrile)copper(I) hexafluorophosphate (Cu(I)-ACN) was dissolved in pure acetonitrile and the amount of Cu(I)-ACN to add to the protein calculated such that the final ACN concentration was 10% of the total protein solution by volume. The Cu(I)-ACN was added to the apo protein anaerobically by syringe pump (1μL/minute rate), with stirring, at a ratio of 1.5:1 metal to protein. The mixture was then allowed to incubate one additional hour with stirring, over ice. The metallated proteins were concentrated to the desired volume using a microconcentrator, and three cycles of desalting were accomplished using spin columns using buffer containing 10%, 5%, and 0% acetonitrile respectively, which removed excess metal and salt from the proteins. The proteins were then flash-frozen in XES sample cuvettes. Metal to protein concentrations were verified by ICP-OES and the BCA or Bradford assay.

X-ray Emission Data Collection

The XES experiments were performed on beamline 6-2 at Stanford Synchrotron Radiation Lightsource with an operating ring current of 500mA. The beamline delivered to the sample spot an incident x-ray beam of 10.5 keV with ~10¹³ photons/s via a liquid N₂ cooled Si(111) monochromator, focused to ~250×700 μm² (FWHM) by means of a Rh-coated Si mirror. The Si(111) monochromator was calibrated using the x-ray absorption of a metallic Cu foil with the first infection point set to 8978.9eV. Then the monochromator energy was scanned through the energy range of the Kβ emission, and the XES spectrometer was calibrated using elastic scattering peaks at each monochromator position. The samples were kept at a temperature of 10K using a liquid He flow cryostat. Multiple spots on each sample were used for collecting an overall (averaged) XES spectra as follows. First a spot in the center of the sample was chosen, and scanned between 8875 and 8930 eV in 0.25 eV steps counted for 1 second per step, to collect a spectrum of the Kβ mainline. Data collection was then switched to the VtC region which was scanned between 8925 and 9020 eV in 1.5 eV steps, counted for 2 seconds per step. The 5 eV overlap between mainline and VtC regions ensured that VtC spectra could be normalized to the intensity of the mainline peak. For VtC spectra, a total of 32 independent spots were measured per sample, with two scans collected per spot for a total of 64 scans per sample. These were subsequently averaged to generate an overall VtC spectrum for each sample. The total dose per exposed spot was below the radiation damage threshold as determined with consecutive XES spectra, and/or XAS time scans monitored using the shape of the absorption edge feature at 8983 eV in a separate absorption scan. This procedure indicated that the fully reduced (Cu(I)) proteins were resistant to radiation damage over the time-course of the XES data collection.

The XES spectra were recorded using the BL 6-2 multicrystal Johann spectrometer³⁷ employing six Si(551) spherical analyzers (100mm in diameter with a 1m radius of curvature) aligned on intersecting Rowland circles. A silicon drift detector was used at the focus of the spectrometer for recording the analyzed photons. A He-filled polyethylene bag was placed between the cryostat and the spectrometer to minimize signal attenuation and diffused scattering contribution from air. The energy resolution of the spectrometer was determined to be of ~1 eV (FWHM) via elastic scattering scans along the energy range used for the Cu VtC XES.

Data Processing

The Cu Kβ XES data were fit using a holistic model that included pseudo-Voigt functions to account for all the Kβ XES features and an offset, which accounted for Kα XES intensity. The differences in effective concentration (due to measurements on different sample spots) were accounted for by the use of a floating parameter in the fit model, which allowed for the alignment of VtC and Kβ mainline scan regions. To better highlight the intensity distributions of the VtC features the Kβ mainline was subtracted and the total area under the VtC region between 8960 eV and 8985 eV was parameterized to reflect 1 unit of normalized intensity. Two spectral features (above 8980 eV) attributed to K+L excitations were included in the model, but the parameterization of their intensities was not included in the total (Kβ+VtC) intensity. Using Blueprint XAS, the solution space and the uncertainty of the fit parameters was explored by obtaining a large family of good fits based on sum of squared errors, as previously described.^38,39 The fitted offset, mainline spectral contributions were then subtracted to better highlight the VtC region for each spectrum. The average for the good fits for all spectra are provided in the supplementary information (Supporting Information S1).

Theoretical Calculations

All theoretical calculations were carried out with the ORCA 3.0 computational chemistry package.⁴⁰ Geometry optimizations were carried out for each metal site using the first coordination sphere amino acid residues from crystal structures where available and a Cu(I) metal center. For CusB, the starting geometry for metal binding site was approximated as trigonal planar using previously reported experimental bond distances and theoretically determined geometry coordination.^20,23 The BP86 functional,^41,42 was used for all calculations. For geometry optimizations the backbone alpha carbons were frozen and the def2-TZVP basis set^43,44 was used for the metal center and directly coordinating atoms, while the def2-SVP was employed for the remaining atoms. Solvation effects were accounted for using a conductor like screening model (COSMO) with a dielectric constant of 30, approximating what is expected at the protein surface.⁴⁵ The geometry-optimized structures were then used to calculate the VtC XES spectra using the def2-TZVP basis set on all atoms. The VtC XES spectra were calculated within the single-point calculation routine using a one-electron theoretical protocol.⁴⁶ In order to facilitate facile comparison with experimental spectra, individual transitions for calculated spectra were described using a Gaussian peak function having a 2.6 eV FWHM and the sum total of the transition intensities in the VtC region was normalized to 1. The overall experimental resolution, accounting for both spectrometer broadening (1 eV) and the Cu-1s core hole lifetime (1.55 eV), is expected to be 1.85 eV. Therefore, the broadening applied to the calculated transitions suggests that the spectral features are not limited by the experimental resolution. Furthermore, correlation of calculated transitions to orbitals as well as model studies to investigate the effect of binding mode were carried out on truncated models of the metal sites with imidazoles and methylthioethers replacing histidines and methionines, respectively. No significant deviation in calculated VtC XES spectra was observed due to the truncation (Supporting Information S6).

Results and Discussion

Previous XAS studies

Before discussing the present XES data it is useful to briefly summarize previous Cu K-edge XAS studies on CusB (Met₃), CusF (Met₂His) and the PHM variants Cu_M (Met₁His₂) and Cu_H (His₃) bound to Cu(I).^9,17,20 In all cases, the XAS spectra are consistent with a three coordinate Cu(I) center. This is supported by both a lack of any pre-edge features attributable to 1s→3d transitions, and a well-resolved feature at ~8983 eV, attributed to 1s→4p transition.⁴⁷ Furthermore, in the case of the PHM Cu_M site, binding of CO leads to a decrease in the intensity of the ~8983 eV feature consistent with an increased coordination number dominated by a nitrogen, oxygen or carbon ligand sphere.⁴⁷ We emphasize, however, that while XAS is a powerful probe of coordination environment and bond metrics, the lack of a pre-edge in d¹⁰ system limits the quantitative information that can be obtained on the local site symmetry. Hence the present study was initiated in order to determine the complementary information that can be obtained by using Cu Kβ XES to directly probe the filled orbital in closed shell d¹⁰ system.

Kβ XES Spectra

Kβ XES spectra have two parts, the mainline (or Kβ_1,3 region) consisting of emission lines from the filled metal 3p orbitals, and the valence to core (VtC or Kβ_2,5+Kβ”) region corresponding to transitions from filled valence orbitals.^48,49 Mainlines can provide insight into the spin state and covalency at the metal center governed by a p-d exchange coupling.⁴⁹ However, in the present case the d¹⁰ configuration at the Cu(I) results in Kβ_1,3 mainlines which are superimposable and do not show any observable spectral changes due to changes in the ligand coordination environment (Fig 2). The Kβ_1,3 mainlines for all proteins in the present series have an intense peak at 8905.9 eV, a shoulder at ~8903.6 eV and an additional weak shoulder at 8897.9 eV.

Figure 2.

Experimental mainline Kβ_1,3 XES spectra and first derivatives highlighting features of interest. (M₁H₅CO) refers to the presence of both the Cu_M (M₁H₂) and Cu_H (H₃) sites with a CO molecule bound to Cu_M. Spectra were normalized by setting the total area under the curve (VtC + mainlines) to 1000.

VtC XES spectra

Figure 3 shows the VtC spectra for CusB (Met₃), CusF (Met₂His) and the PHM variants Cu_M (Met₁His₂) and Cu_H (His₃). All spectra exhibit features in the Kβ_2,5 region (~ 8965 – 8980 eV), which are usually associated with transitions from ligand np orbitals to the 1s core hole of the metal. The energy positions for VtC features are generally correlated with the ligand ionization potential, while the intensity is highly dependent on metal-ligand bond distance.^46,50 Hence, the systematic variation in ligand speciation for the present series should provide insights into the nature of histidine versus methionine metal coordination. In particular, clear differences are observed in the VtC region at the Met₃ and His₃ limits. CusB (Met₃) has a distinct peak at 8978.4 eV with a broad shoulder at 8975.6 eV. Cu_H (His₃) has two well-resolved features: a more intense feature at 8974.8 eV and a second at 8977.2 eV. The mixed coordination sphere CusF (Met₂His) and the PHM Cu_M variant (Met₁His₂) have very similar spectra consisting of two features at 8975.2 eV and 8977.8 eV. The picture becomes more complex in mixed coordination spheres, as both the nature of the ligand as well as variations in metal-ligand distance are known to influence the spectra.^31,39,46,51 Figure 3 (right) also shows the calculated VtC XES spectra, for the series of Cu proteins. The trends from the calculations generally agree well with experiment, with the exception of CusF (Met₂His), where the intensity of the high energy VtC feature seems to be overestimated. To help understand the contributions from the methionine and histidines to the VtC spectra, the principal calculated transitions to the CusB and Cu_H spectra were further analyzed.

Figure 3.

(left) Experimental VtC XES spectra; (right) Calculated VtC XES spectra. Calculated spectra were shifted by 232 eV.

Theoretical assignment of VtC XES spectra

In order to simplify the orbital picture, hisitidnes were approximated as imidazoles and methionines as methylthioethers. No significant deviation in calculated VtC XES spectra was observed due to the truncation. Figure 4 shows an overlay of the calculated and experimental spectra for the trigonal planar CusB Met₃ site. A breakdown of the transitions that contribute to the VtC region reveals four primary sets of transitions, as shown in Figure 4 (bottom). As the calculation uses a one-electron approximation, the transitions can be correlated with molecular orbitals.^39,46 The two most intense features, which appear at ~8978.4 eV and at ~8975.6 eV correspond to transitions from the antibonding (1) and bonding (3 and 4) combinations of the d_x²−y² and d_xy orbitals with the S 3p of the methionines, respectively. Two distinct sets of S 3p orbitals are involved, a lone pair S 3p orbital (3) and a S 3p orbital from the S-CH_3σ bond (4). Transitions from these orbitals gain intensity due to the relatively high percentage of Cu p character present in these molecular orbitals. In D_3h symmetry the d_x²−y² and d_xy orbitals have the appropriate symmetry to mix with metal p orbitals; and in a configuration interaction based model, these transitions can gain intensity through covalent ligand interactions, which serve as an intermediary to promote p-d mixing in appropriate symmetry.⁵² In this context, one notes that both the Cu p and S p character in these orbitals are significant (Figure 4 - middle). In contrast, transitions from the antibonding and bonding combination with the d_xz, d_yz and d_z² set (2), which are not of appropriate symmetry to interact with the metal p orbitals do not have significant intensity.

Figure 4.

(top) Overlay of the experimental and calculated Kβ_2,5 VtC XES for CusB. The calculated spectra were shifted by 232 eV and individual transitions are shown as sticks. The truncated model was used. (middle) Percent character breakdown for individual transitions. (bottom) Selected molecular orbitals corresponding to transitions of interest (isovalue = 0.05).

Similarly for Cu_H, which has a coordination sphere consisting of three histidines (imidazoles), the calculated spectrum accurately predicts the two experimentally observed features (Figure 5). Again the two most intense transitions (1' and 4') correspond to transitions from the antibonding and bonding combinations, respectively, of the d_x²−y² and d_xy orbitals with the ligand. We note however, that the imidazole character in the d_x²−y² and d_xy orbitals is much less (~14%) than for the methionine (thioether) ligands (~35%) resulting in less p-d mixing and lower overall transition intensities. In contrast to methionine, the most intense set of transitions in the Cu_H VtC spectra (4') come from the N-(CH₂)_2σ -bonding orbitals. These form a σ-bonding interaction with the Cu (and a corresponding antibonding interaction in 1'). The equivalent of the S p lone pair for the metal bonded N in imidazole (3') does not interact strongly with the metal, as it forms part of the aromatic π manifold of the imidazole ring (Figure 5, bottom). Furthermore, the weak metal-ligand interaction that is present is a π-type interaction, which does not favor mixing with Cu p-orbitals, and as previously described, results in further loss of VtC intensity.^39,48

Figure 5.

(top) Overlay of the experimental and calculated Kβ_2,5 VtC XES for Cu_H. The calculated spectra were shifted by 232 eV and individual transitions are shown as sticks. The truncated model was used. (middle) Percent character breakdown for individual transitions. (bottom) Selected molecular orbitals corresponding to transitions of interest (isovalue = 0.05).

Effect of ligand geometry on the lineshape of Cu(I) VtC XES spectra

Generally, for VtC XES the transition intensities and energies have been correlated with two principal factors: the metal-ligand bond distance and the ligand ionization potential.^28,46,50 For the former, the shorter the bond distance the more metal character np can mix into the orbital, resulting in a more intense transition. For the latter, the higher the ionization potential of the ligand the lower in energy the corresponding VtC features appear. However, in the current study, the higher energy VtC features arise from filled metal d-orbitals, which have ligand-mediated p mixing. Although the metal d-manifold is not generally expected to contribute significantly to the VtC spectra, VtC features from metal d orbitals have previously been reported in iron-carbonyl complexes and more recently in a series of manganese dioxygen activating small molecules.^39,53 This effect is expected to become more pronounced, as in the current case, when a filled d¹⁰ shell is present. Therefore a third factor also affects the Cu(I) VtC lineshape: ligand-mediated metal p-d mixing.

To further investigate this factor, and in turn to help potentially explain some of the discrepancies in the calculated and experimental spectra, variations in the ligand geometry and their effect on the VtC XES were explored using model systems. A simplified model of the CusF metal binding site consisting of an imidazole and two thioethers was built and the geometry was allowed to relax resulting in a S-Cu-S angle of 116°. Variation of the S-Cu-S angle was found to result in markedly different “Cu d” VtC features (Figure 6). This can be rationalized in terms of ligand mediated Cu p-d mixing, dependent on the overlap between the S p-type orbitals with the Cu d manifold. In the case of a S-Cu-S angle of 116°, the d_x²−y² orbital interacts predominantly with the histidine, however, some methionine character is also present. The methionine ligands, on the other hand, interact more directly with the d_xy orbital. A contraction of the S-Cu-S angle is predicted to reduce the amount of methionine sulfur mixing into the d_x²−y² orbital, which in turn significantly lowers the amount of Cu p mixing (Figure 6 - bottom). The d_xy orbital, pointing at the methionine ligands on the other hand, is largely unaffected. The result is a decrease in intensity of the feature at 8977.8 eV, which is more consistent with the experimental data. Crystallographic analysis of the apo-, holo-Ag(I) and holo-Cu(I) CusF binding site, shows flexibility in the orientation of the ligand residues and backbone particularly the methionines, which are the most solvent exposed residues in the metal binding site.^17,19 Therefore a contraction of the S-Cu-S angle is certainly one possible origin of the discrepancy between the experimental and calculated spectra for CusF. Modulation of spectral intensities by varying metal-ligand bond distance and geometry were also investigated, but were found not to correlate with the overall experimental spectral line-shape (Supporting Information S7).

Figure 6.

In silico study of the effect on the CusF VtC XES spectra of S-Cu-S angle variations (top); relative orientation of the d_x²−y² orbital with respect to the S ligands (inset); percent Cu d and p character mixed in the d_xy and d_x²−y² orbitals (bottom). Calculated spectra were shifted by 232 eV.

Variation of p-d mixing via coordination geometry was also investigated in the Cu_H site (Figure 7). In particular, a comparison of a T-shape and trigonal planar geometry for the His₃ site was carried out. Crystallographic analysis indicates a distorted T-shape geometry for the Cu_H site.^3,4 At the T-shape coordination limit, the imidazole ligands interact predominantly with the d_x²−y² orbital, increasing the amount of p mixing and increasing the energy of the d_x²−y² orbital. At the same time, ligand interaction with the d_xy orbital is minimized. The overall result is a drop in intensity and a shift to higher energy for the transition at 8977.2 eV, which is in contrast to what one observes in the experimental data. A theoretical model more consistent with experimental data is that of a trigonal planar geometry, where both calculated features of the Cu_H site match experimental spectra. Therefore, in solution, the Cu_H site may be better described as having a more trigonal planar coordination geometry rather than a T-shape. This is also more consistent with preferred Cu(I) geometry.¹⁹

Figure 7.

In silico study of the effect on the Cu_H VtC XES spectra of T-shape, trigonal planar and trigonal pyramidal geometries (top); relative orientation of the d_x²−y² orbital with respect to the ligands (inset); percent Cu d and p character mixed in the d_xy and d_x²−y² orbitals (bottom). Calculated spectra were shifted by 232 eV.

The above discussion highlights the possibility of using VtC XES, for electron rich metal centers such as d¹⁰ systems, to gain insights into their coordination environment, both in terms of ligand speciation and site symmetry. This can complement sister techniques such as X-ray absorption, where it is often difficult to extract coordination geometry information for metal centers with filled d shells. This is due to the general lack of pre-edge features (1s → 3d) and rising edge features (1s → 4p) that not only depend on geometry but covalency arguments as well.^47,54,55

PHM-CO VtC XES spectra

With an understanding of the factors affecting the spectral line-shape, the applicability of VtC XES for exploring the PHM mechanisms was tested using CO bound WT PHM (PHM-CO). Figure 8 shows the experimental spectra of PHM-CO superimposed with that of PHM. The spectrum of PHM was generated by averaging the spectra of the Cu_M site and Cu_H site. Two distinct differences are observed in the experimental spectra. PHM-CO has a more intense feature at 8978.3 eV attributed to transitions from the Cu d-manifold and a shoulder at 8969.5 eV, presumably due to the bound CO moiety. Calculated spectra of PHM and PHM-CO generally agree well with the experimental data (Figure 8-right).

Figure 8.

VtC XES spectra of PHM and PHM-CO; (left) Experimental; (right). (M₁H₅) refers to the presence of both the Cu_M (M₁H₂) and Cu_H (H₃) sites Calculated spectra were shifted by 232 eV.

To highlight the changes upon CO binding, both the experimental and calculated difference spectra of PHM-CO and PHM are overlaid in Figure 9. The Cu_M site of PHM-CO is a distorted tetrahedral. Crystallographic data highlights an elongation of the Cu-S bond for CO bound Cu_M. Presumably this helps maintain the site closer to the trigonal planar geometry favored by Cu(I), and similar behavior has previously been reported in proteins such as plastocyanin.^19,56 Nevertheless a distorted tetrahedral model is consistent with the experimental data. The increase in intensity in the Cu d region (5) of the VtC can be explained by the covalent interaction between the copper center and the CO ligand, resulting in more p-d mixing. Furthermore two additional features at ~8970 eV (7) and ~8972.5 eV (6) are present and are consistent with previously reported VtC spectra of Fe-CO complexes (Figure 9). Complexes such as [Fe(CO)₅] are reported to have two distinct features due to the CO moiety separated by ~ 3 eV. The lower energy feature was assigned to metal interactions with the σ*_2s-2s CO anti-bonding orbital, while the higher energy feature is attributed to the CO σ_2p-2p.³⁹ The current result highlights that VtC XES is a suitable approach to detected small molecule interactions in PHM despite the presence of two non-equivalent Cu(I) sites.

Figure 9.

(top) Overlay of the experimental and calculated VtC XES difference spectra of PHM-CO and PHM. The calculated spectra were shifted by 232 eV. The truncated model was used. (middle) Percent character breakdown for individual transitions. (bottom) Molecular orbitals corresponding to features of interest (isovalue = 0.05).

Summary and Conclusion

Herein a series of Cu(I) metal binding sites with a systematic variation in ligand speciation (His/Met) were investigated with VtC XES. In all cases, a high energy feature at ~8977 eV, attributed to transitions arising from filled Cu 3d orbitals was observed. Computational analysis of this region demonstrated that the intensities of the Cu 3d transition are modulated by ligand mediated Cu p-d mixing, through a configuration interaction mechanism. Sulfur-based ligands were found to more readily mix with the d-manifold resulting in more intense transitions in this region when methionines were present than when histidines were present. This is facilitated by the presence of two Cu-ligand interactions with methionines: one from the S 3p lone pair, and one from the S-CH₃σ bond. In addition, both Cu(I) 3d transition energies and intensities can be modulated by the geometry around the metal and metal-ligand binding mode. Variations in geometry and/or binding mode of the ligand affect the overlap between the ligand and the d-orbitals, altering the p-d mixing and thus the observed transition intensities. Therefore, in the case of electron rich metal centers such as those having filled d¹⁰ shells, the VtC region might serve as a probe of not only ligand speciation but also coordination geometry, in a fashion similar to XAS pre-edges. Therefore VtC XES could be used in conjunction with complementary techniques such as X-ray absorption to offer a more complete picture of the environment around the metal center. Finally, it was observed that VtC XES is sensitive to small molecule coordination in PHM, suggesting it would be an appropriate method for future mechanistic studies.